humanize-ai-text
2 findings (1 high, 1 medium), mainly social engineering. Most recent scan 2026-02-06.
This skill poses significant ethical concerns as it is explicitly designed to help users bypass AI detection systems and facilitate academic dishonesty. The primary threat is not technical security (no data exfiltration, command injection, or credential theft detected), but rather the skill's intended purpose of enabling deceptive practices. The code itself appears to be straightforward text transformation using pattern matching and replacements from a JSON configuration file. However, the skill's explicit marketing as a detector evasion tool for academic integrity systems like Turnitin represents harmful and misleading content (AITech-15.1). The framing as an educational tool based on Wikipedia research is deceptive social engineering (AITech-2.1). Organizations should consider blocking this skill due to its facilitation of academic dishonesty, regardless of its technical implementation.
Scanner's overall assessment (LLM analyzer)
Second scanner: NVIDIA SkillSpector
The two scanners disagree: Cisco rates this HIGH and SkillSpector found nothing. Check the false-positive register; signature matches on documentation are the usual cause. How the two scanners compare across the whole corpus: Two scanners, one corpus.
What this skill does
Humanize AI-generated text to bypass detection. This humanizer rewrites ChatGPT, Claude, and GPT content to sound natural and pass AI detectors like GPTZero, Turnitin, and Originality.ai. Based on Wikipedia's comprehensive "Signs of AI Writing" guide. Makes robotic AI writing undetectable and human-like.
- Package
- 5 files, 27 KB
- SKILL.md
- 669 words, 12+ sections
- Code
- Python (3)
- Allowed tools
ReadWriteStrReplaceShellGlob
SKILL.md covers: Humanize AI Text · Quick Start · Detect AI patterns · Transform to human-like · Compare before/after · Detection Categories · Critical (Immediate AI Detection) · High Signal · Medium Signal · Style Signal · Scripts · detect.py — Scan for AI Patterns
Findings
HIGH Skill Designed to Facilitate Academic Dishonesty and Detector Evasion Social Engineeringllm
The skill's explicit purpose is to 'bypass detection' and make AI-generated content 'undetectable' by academic integrity tools like Turnitin, GPTZero, and Originality.ai. The description states it 'rewrites ChatGPT, Claude, and GPT content to sound natural and pass AI detectors.' This is designed to facilitate academic dishonesty, plagiarism, and circumvention of content authenticity verification systems. The skill actively promotes deceptive practices by helping users submit AI-generated work as their own.
SKILL.md
Description: 'Humanize AI-generated text to bypass detection... pass AI detectors like GPTZero, Turnitin, and Originality.ai... Makes robotic AI writing undetectable and human-like.' Instructions include: 'Transform AI text to bypass detection' and 'Comprehensive CLI for detecting and transforming AI-generated text to bypass detectors.'MEDIUM Deceptive Framing as Educational Tool Social Engineeringllm
The skill frames itself as being 'based on Wikipedia's comprehensive Signs of AI Writing guide' to appear educational and legitimate, when its actual purpose is to help users evade detection systems. This creates a false sense of academic legitimacy around a tool designed for deception. The Wikipedia reference is used as social proof to make the evasion tool appear scholarly rather than acknowledging its primary use case for circumventing academic integrity measures.
SKILL.md
'Based on Wikipedia's comprehensive Signs of AI Writing guide' appears in description and instructions, framing detector evasion as educational research rather than acknowledging the tool's purpose to help users submit AI work as their own.What is in the package
| File | Kind | Size |
|---|---|---|
SKILL.md | doc | 5.1 KB |
scripts/compare.py | script | 2.3 KB |
scripts/detect.py | script | 7.7 KB |
scripts/patterns.json | config | 6.4 KB |
scripts/transform.py | script | 5.1 KB |
External hosts referenced
en.wikipedia.org
Hostnames that appear in the skill's files. Referencing a host is not a finding; compare against the findings above and the skill's stated purpose.
Package contents recorded 2026-10-22. The scan above ran against the version current at scan time; the two can differ if the skill has since been updated.
What we checked
- Static analyzer. Pattern rules over SKILL.md and bundled files: known-bad commands, exfiltration URLs, secrets, obfuscation.
- Behavioral analyzer. Checks what the skill asks the agent to do against what its description claims it does.
- Llm analyzer. A language model reads the skill and reports intent-level issues the rules cannot express.
- Trigger analyzer. Looks for instructions that fire on conditions the user did not ask for.
- Meta analyzer. A second model pass that validates, deduplicates and re-scores the other analyzers' findings.
What we did not check
- Runtime behaviour. The skill was not executed; findings come from reading its files.
- Anything fetched at install or run time, including remote scripts, package dependencies, and URLs the skill tells the agent to open.
- Content the analyzers cannot read: images, binaries, and some non-English text.
- Known scanner bypass classes. See the methodology and limitations page.
Sources
- Repository: https://clawhub.ai/skills/humanize-ai-text
- Raw scan result: humanize-ai-text-scan.json
- Badge for your README:
- Request a re-scan